lib-MOCS-KMC364-20140106084043


211 

NAME-TITLE ENTRY RETRIEVAL FROM A MARC FILE 

Philip L. LONG, Head, Automated Systems Research and Development 
and Frederick G. KILGOUR, Director: Ohio College Library Center, 
Columbus, Ohio 

A test of validity of earlier findings on 3,3 search-ke y retrieval from an 
in-process file for retrieval from a MARC file. Probability of number of 
entries retrieved per reply is essentially the same for both files. 

This study was undertaken to test the applicability of previous findings 
on retrieval of name-title entries from a technical processing system fil e 
( 1 ) to retrieval from a MARC file; the technique for retrieval employs 
truncated 3,3 search keys. 

MATERIALS AND METHODS 
The study cited above employed a file of 132,808 name-title entries 

obtained from the Yale University Library's Machine Aided Technical 
Processing System. Bibliographic control was not maintained for the 
generation of records in this file , with the result that the file contained 
errors that simulated errors in the requests library users put to catalogs. 
The MARC file employed in the present study contains 121,588 name-title 
entries that are nearly error free. Whereas the MARC file possesses few 
records bearing foreign titles, the Yale file has a significantly higher per-
centage of such titles, as would be expected for a large university library. 
Initial articles were deleted in Yale titles, but only English articles in 
MARC titles because the language of foreign language titles is not identified 
in MARC. 



212 Journal of Library Automation Vol. 4/4 December, 1971 

Design of the program used to analyze the MARC file was the same as 
that for the program employed in the previous study. However, the new 
program runs on a Xerox Data Systems Sigma 5 computer. The test em-
ployed the 3,3 search key to make possible comparison with previous 
results. 

RESULTS 

Table 1 presents the percentage of time that up to five replies can be 
expected, assuming equal likelihood of key choice. Inspection of the table 
reveals that there is no significant difference between the findings from 
the Yale and the MARC files. 

Table 1. Probability of Number of Entries Per Reply Using 3,3, Search Key 
Number of Replies 

1 
2 
3 
4 
5 

DISCUSSION 

Cumulative Probability Percentage 
Yale File MARC File 
78.58 79.98 
92.75 93.28 
96.83 96.93 
98.40 98.26 
99.08 98.91 

The same result was expected for the MARC file that had been obtained 
earlier from the Yale file. Possible influences that might have led to 
different results were the existence of errors in the Yale file, a significant 
proportion of foreign titles in the Yale file as compared to the nearly 
all-English MARC file, and the inability to mechanically delete the initial 
articles in the few foreign language MARC titles. It is most unlikely that 
the effects of these differences are masking one another. 

CONCLUSION 

The findings of a previous study on the effectiveness of retrieval of 
entries from a large bibliographic file ( 1) by use of a truncated 3,3 search 
key have been confirmed for a similarly large MARC file. 

REFERENCE 

1. Kilgour, Frederick G.; Long, Philip L. ; Leiderman, Eugene B.: "Retrieval 
of Bibliographic Entries from a Name-Title Catalog by Use of Truncated 
Search Keys," Proceedings of the American Society for Information 
Science, 7 ( 1970 ), 79-81. 

-